[net-next-2.6.git] / fs / nfsd / nfssvc.c

/*
 * linux/fs/nfsd/nfssvc.c
 *
 * Central processing for nfsd.
 *
 * Authors:	Olaf Kirch (okir@monad.swb.de)
 *
 * Copyright (C) 1995, 1996, 1997 Olaf Kirch <okir@monad.swb.de>
 */

#include <linux/module.h>
#include <linux/sched.h>
#include <linux/time.h>
#include <linux/errno.h>
#include <linux/nfs.h>
#include <linux/in.h>
#include <linux/uio.h>
#include <linux/unistd.h>
#include <linux/slab.h>
#include <linux/smp.h>
#include <linux/smp_lock.h>
#include <linux/freezer.h>
#include <linux/fs_struct.h>
#include <linux/kthread.h>

#include <linux/sunrpc/types.h>
#include <linux/sunrpc/stats.h>
#include <linux/sunrpc/svc.h>
#include <linux/sunrpc/svcsock.h>
#include <linux/sunrpc/cache.h>
#include <linux/nfsd/nfsd.h>
#include <linux/nfsd/stats.h>
#include <linux/nfsd/cache.h>
#include <linux/nfsd/syscall.h>
#include <linux/lockd/bind.h>
#include <linux/nfsacl.h>

#define NFSDDBG_FACILITY	NFSDDBG_SVC

/* these signals will be delivered to an nfsd thread 
 * when handling a request
 */
#define ALLOWED_SIGS	(sigmask(SIGKILL))
/* these signals will be delivered to an nfsd thread
 * when not handling a request. i.e. when waiting
 */
#define SHUTDOWN_SIGS	(sigmask(SIGKILL) | sigmask(SIGHUP) | sigmask(SIGINT) | sigmask(SIGQUIT))

extern struct svc_program	nfsd_program;
static int			nfsd(void *vrqstp);
struct timeval			nfssvc_boot;
static atomic_t			nfsd_busy;
static unsigned long		nfsd_last_call;
static DEFINE_SPINLOCK(nfsd_call_lock);

/*
 * nfsd_mutex protects nfsd_serv -- both the pointer itself and the members
 * of the svc_serv struct. In particular, ->sv_nrthreads but also to some
 * extent ->sv_temp_socks and ->sv_permsocks. It also protects nfsdstats.th_cnt
 *
 * If (out side the lock) nfsd_serv is non-NULL, then it must point to a
 * properly initialised 'struct svc_serv' with ->sv_nrthreads > 0. That number
 * of nfsd threads must exist and each must listed in ->sp_all_threads in each
 * entry of ->sv_pools[].
 *
 * Transitions of the thread count between zero and non-zero are of particular
 * interest since the svc_serv needs to be created and initialized at that
 * point, or freed.
 *
 * Finally, the nfsd_mutex also protects some of the global variables that are
 * accessed when nfsd starts and that are settable via the write_* routines in
 * nfsctl.c. In particular:
 *
 *	user_recovery_dirname
 *	user_lease_time
 *	nfsd_versions
 */
DEFINE_MUTEX(nfsd_mutex);
struct svc_serv 		*nfsd_serv;

#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
static struct svc_stat	nfsd_acl_svcstats;
static struct svc_version *	nfsd_acl_version[] = {
	[2] = &nfsd_acl_version2,
	[3] = &nfsd_acl_version3,
};

#define NFSD_ACL_MINVERS            2
#define NFSD_ACL_NRVERS		ARRAY_SIZE(nfsd_acl_version)
static struct svc_version *nfsd_acl_versions[NFSD_ACL_NRVERS];

static struct svc_program	nfsd_acl_program = {
	.pg_prog		= NFS_ACL_PROGRAM,
	.pg_nvers		= NFSD_ACL_NRVERS,
	.pg_vers		= nfsd_acl_versions,
	.pg_name		= "nfsacl",
	.pg_class		= "nfsd",
	.pg_stats		= &nfsd_acl_svcstats,
	.pg_authenticate	= &svc_set_client,
};

static struct svc_stat	nfsd_acl_svcstats = {
	.program	= &nfsd_acl_program,
};
#endif /* defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL) */

static struct svc_version *	nfsd_version[] = {
	[2] = &nfsd_version2,
#if defined(CONFIG_NFSD_V3)
	[3] = &nfsd_version3,
#endif
#if defined(CONFIG_NFSD_V4)
	[4] = &nfsd_version4,
#endif
};

#define NFSD_MINVERS    	2
#define NFSD_NRVERS		ARRAY_SIZE(nfsd_version)
static struct svc_version *nfsd_versions[NFSD_NRVERS];

struct svc_program		nfsd_program = {
#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
	.pg_next		= &nfsd_acl_program,
#endif
	.pg_prog		= NFS_PROGRAM,		/* program number */
	.pg_nvers		= NFSD_NRVERS,		/* nr of entries in nfsd_version */
	.pg_vers		= nfsd_versions,	/* version table */
	.pg_name		= "nfsd",		/* program name */
	.pg_class		= "nfsd",		/* authentication class */
	.pg_stats		= &nfsd_svcstats,	/* version table */
	.pg_authenticate	= &svc_set_client,	/* export authentication */

};

int nfsd_vers(int vers, enum vers_op change)
{
	if (vers < NFSD_MINVERS || vers >= NFSD_NRVERS)
		return -1;
	switch(change) {
	case NFSD_SET:
		nfsd_versions[vers] = nfsd_version[vers];
#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
		if (vers < NFSD_ACL_NRVERS)
			nfsd_acl_versions[vers] = nfsd_acl_version[vers];
#endif
		break;
	case NFSD_CLEAR:
		nfsd_versions[vers] = NULL;
#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
		if (vers < NFSD_ACL_NRVERS)
			nfsd_acl_versions[vers] = NULL;
#endif
		break;
	case NFSD_TEST:
		return nfsd_versions[vers] != NULL;
	case NFSD_AVAIL:
		return nfsd_version[vers] != NULL;
	}
	return 0;
}
/*
 * Maximum number of nfsd processes
 */
#define	NFSD_MAXSERVS		8192

int nfsd_nrthreads(void)
{
	if (nfsd_serv == NULL)
		return 0;
	else
		return nfsd_serv->sv_nrthreads;
}

static void nfsd_last_thread(struct svc_serv *serv)
{
	/* When last nfsd thread exits we need to do some clean-up */
	struct svc_xprt *xprt;
	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list)
		lockd_down();
	nfsd_serv = NULL;
	nfsd_racache_shutdown();
	nfs4_state_shutdown();

	printk(KERN_WARNING "nfsd: last server has exited, flushing export "
			    "cache\n");
	nfsd_export_flush();
}

void nfsd_reset_versions(void)
{
	int found_one = 0;
	int i;

	for (i = NFSD_MINVERS; i < NFSD_NRVERS; i++) {
		if (nfsd_program.pg_vers[i])
			found_one = 1;
	}

	if (!found_one) {
		for (i = NFSD_MINVERS; i < NFSD_NRVERS; i++)
			nfsd_program.pg_vers[i] = nfsd_version[i];
#if defined(CONFIG_NFSD_V2_ACL) || defined(CONFIG_NFSD_V3_ACL)
		for (i = NFSD_ACL_MINVERS; i < NFSD_ACL_NRVERS; i++)
			nfsd_acl_program.pg_vers[i] =
				nfsd_acl_version[i];
#endif
	}
}


int nfsd_create_serv(void)
{
	int err = 0;

	WARN_ON(!mutex_is_locked(&nfsd_mutex));
	if (nfsd_serv) {
		svc_get(nfsd_serv);
		return 0;
	}
	if (nfsd_max_blksize == 0) {
		/* choose a suitable default */
		struct sysinfo i;
		si_meminfo(&i);
		/* Aim for 1/4096 of memory per thread
		 * This gives 1MB on 4Gig machines
		 * But only uses 32K on 128M machines.
		 * Bottom out at 8K on 32M and smaller.
		 * Of course, this is only a default.
		 */
		nfsd_max_blksize = NFSSVC_MAXBLKSIZE;
		i.totalram <<= PAGE_SHIFT - 12;
		while (nfsd_max_blksize > i.totalram &&
		       nfsd_max_blksize >= 8*1024*2)
			nfsd_max_blksize /= 2;
	}

	atomic_set(&nfsd_busy, 0);
	nfsd_serv = svc_create_pooled(&nfsd_program, nfsd_max_blksize,
				      nfsd_last_thread, nfsd, THIS_MODULE);
	if (nfsd_serv == NULL)
		err = -ENOMEM;

	do_gettimeofday(&nfssvc_boot);		/* record boot time */
	return err;
}

static int nfsd_init_socks(int port)
{
	int error;
	if (!list_empty(&nfsd_serv->sv_permsocks))
		return 0;

	error = lockd_up(IPPROTO_UDP);
	if (error >= 0) {
		error = svc_create_xprt(nfsd_serv, "udp", port,
					SVC_SOCK_DEFAULTS);
		if (error < 0)
			lockd_down();
	}
	if (error < 0)
		return error;

	error = lockd_up(IPPROTO_TCP);
	if (error >= 0) {
		error = svc_create_xprt(nfsd_serv, "tcp", port,
					SVC_SOCK_DEFAULTS);
		if (error < 0)
			lockd_down();
	}
	if (error < 0)
		return error;
	return 0;
}

int nfsd_nrpools(void)
{
	if (nfsd_serv == NULL)
		return 0;
	else
		return nfsd_serv->sv_nrpools;
}

int nfsd_get_nrthreads(int n, int *nthreads)
{
	int i = 0;

	if (nfsd_serv != NULL) {
		for (i = 0; i < nfsd_serv->sv_nrpools && i < n; i++)
			nthreads[i] = nfsd_serv->sv_pools[i].sp_nrthreads;
	}

	return 0;
}

int nfsd_set_nrthreads(int n, int *nthreads)
{
	int i = 0;
	int tot = 0;
	int err = 0;

	WARN_ON(!mutex_is_locked(&nfsd_mutex));

	if (nfsd_serv == NULL || n <= 0)
		return 0;

	if (n > nfsd_serv->sv_nrpools)
		n = nfsd_serv->sv_nrpools;

	/* enforce a global maximum number of threads */
	tot = 0;
	for (i = 0; i < n; i++) {
		if (nthreads[i] > NFSD_MAXSERVS)
			nthreads[i] = NFSD_MAXSERVS;
		tot += nthreads[i];
	}
	if (tot > NFSD_MAXSERVS) {
		/* total too large: scale down requested numbers */
		for (i = 0; i < n && tot > 0; i++) {
		    	int new = nthreads[i] * NFSD_MAXSERVS / tot;
			tot -= (nthreads[i] - new);
			nthreads[i] = new;
		}
		for (i = 0; i < n && tot > 0; i++) {
			nthreads[i]--;
			tot--;
		}
	}

	/*
	 * There must always be a thread in pool 0; the admin
	 * can't shut down NFS completely using pool_threads.
	 */
	if (nthreads[0] == 0)
		nthreads[0] = 1;

	/* apply the new numbers */
	svc_get(nfsd_serv);
	for (i = 0; i < n; i++) {
		err = svc_set_num_threads(nfsd_serv, &nfsd_serv->sv_pools[i],
				    	  nthreads[i]);
		if (err)
			break;
	}
	svc_destroy(nfsd_serv);

	return err;
}

int
nfsd_svc(unsigned short port, int nrservs)
{
	int	error;

	mutex_lock(&nfsd_mutex);
	dprintk("nfsd: creating service\n");
	error = -EINVAL;
	if (nrservs <= 0)
		nrservs = 0;
	if (nrservs > NFSD_MAXSERVS)
		nrservs = NFSD_MAXSERVS;
	
	/* Readahead param cache - will no-op if it already exists */
	error =	nfsd_racache_init(2*nrservs);
	if (error<0)
		goto out;
	nfs4_state_start();

	nfsd_reset_versions();

	error = nfsd_create_serv();

	if (error)
		goto out;
	error = nfsd_init_socks(port);
	if (error)
		goto failure;

	error = svc_set_num_threads(nfsd_serv, NULL, nrservs);
 failure:
	svc_destroy(nfsd_serv);		/* Release server */
 out:
	mutex_unlock(&nfsd_mutex);
	return error;
}

static inline void
update_thread_usage(int busy_threads)
{
	unsigned long prev_call;
	unsigned long diff;
	int decile;

	spin_lock(&nfsd_call_lock);
	prev_call = nfsd_last_call;
	nfsd_last_call = jiffies;
	decile = busy_threads*10/nfsdstats.th_cnt;
	if (decile>0 && decile <= 10) {
		diff = nfsd_last_call - prev_call;
		if ( (nfsdstats.th_usage[decile-1] += diff) >= NFSD_USAGE_WRAP)
			nfsdstats.th_usage[decile-1] -= NFSD_USAGE_WRAP;
		if (decile == 10)
			nfsdstats.th_fullcnt++;
	}
	spin_unlock(&nfsd_call_lock);
}

/*
 * This is the NFS server kernel thread
 */
static int
nfsd(void *vrqstp)
{
	struct svc_rqst *rqstp = (struct svc_rqst *) vrqstp;
	struct fs_struct *fsp;
	sigset_t shutdown_mask, allowed_mask;
	int err, preverr = 0;
	unsigned int signo;

	/* Lock module and set up kernel thread */
	mutex_lock(&nfsd_mutex);

	/* At this point, the thread shares current->fs
	 * with the init process. We need to create files with a
	 * umask of 0 instead of init's umask. */
	fsp = copy_fs_struct(current->fs);
	if (!fsp) {
		printk("Unable to start nfsd thread: out of memory\n");
		goto out;
	}
	exit_fs(current);
	current->fs = fsp;
	current->fs->umask = 0;

	siginitsetinv(&shutdown_mask, SHUTDOWN_SIGS);
	siginitsetinv(&allowed_mask, ALLOWED_SIGS);

	/*
	 * thread is spawned with all signals set to SIG_IGN, re-enable
	 * the ones that matter
	 */
	for (signo = 1; signo <= _NSIG; signo++) {
		if (!sigismember(&shutdown_mask, signo))
			allow_signal(signo);
	}

	nfsdstats.th_cnt++;
	mutex_unlock(&nfsd_mutex);

	/*
	 * We want less throttling in balance_dirty_pages() so that nfs to
	 * localhost doesn't cause nfsd to lock up due to all the client's
	 * dirty pages.
	 */
	current->flags |= PF_LESS_THROTTLE;
	set_freezable();

	/*
	 * The main request loop
	 */
	for (;;) {
		/* Block all but the shutdown signals */
		sigprocmask(SIG_SETMASK, &shutdown_mask, NULL);

		/*
		 * Find a socket with data available and call its
		 * recvfrom routine.
		 */
		while ((err = svc_recv(rqstp, 60*60*HZ)) == -EAGAIN)
			;
		if (err == -EINTR)
			break;
		else if (err < 0) {
			if (err != preverr) {
				printk(KERN_WARNING "%s: unexpected error "
					"from svc_recv (%d)\n", __func__, -err);
				preverr = err;
			}
			schedule_timeout_uninterruptible(HZ);
			continue;
		}

		update_thread_usage(atomic_read(&nfsd_busy));
		atomic_inc(&nfsd_busy);

		/* Lock the export hash tables for reading. */
		exp_readlock();

		/* Process request with signals blocked. */
		sigprocmask(SIG_SETMASK, &allowed_mask, NULL);

		svc_process(rqstp);

		/* Unlock export hash tables */
		exp_readunlock();
		update_thread_usage(atomic_read(&nfsd_busy));
		atomic_dec(&nfsd_busy);
	}

	/* Clear signals before calling svc_exit_thread() */
	flush_signals(current);

	mutex_lock(&nfsd_mutex);
	nfsdstats.th_cnt --;

out:
	/* Release the thread */
	svc_exit_thread(rqstp);

	/* Release module */
	mutex_unlock(&nfsd_mutex);
	module_put_and_exit(0);
	return 0;
}

static __be32 map_new_errors(u32 vers, __be32 nfserr)
{
	if (nfserr == nfserr_jukebox && vers == 2)
		return nfserr_dropit;
	if (nfserr == nfserr_wrongsec && vers < 4)
		return nfserr_acces;
	return nfserr;
}

int
nfsd_dispatch(struct svc_rqst *rqstp, __be32 *statp)
{
	struct svc_procedure	*proc;
	kxdrproc_t		xdr;
	__be32			nfserr;
	__be32			*nfserrp;

	dprintk("nfsd_dispatch: vers %d proc %d\n",
				rqstp->rq_vers, rqstp->rq_proc);
	proc = rqstp->rq_procinfo;

	/* Check whether we have this call in the cache. */
	switch (nfsd_cache_lookup(rqstp, proc->pc_cachetype)) {
	case RC_INTR:
	case RC_DROPIT:
		return 0;
	case RC_REPLY:
		return 1;
	case RC_DOIT:;
		/* do it */
	}

	/* Decode arguments */
	xdr = proc->pc_decode;
	if (xdr && !xdr(rqstp, (__be32*)rqstp->rq_arg.head[0].iov_base,
			rqstp->rq_argp)) {
		dprintk("nfsd: failed to decode arguments!\n");
		nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
		*statp = rpc_garbage_args;
		return 1;
	}

	/* need to grab the location to store the status, as
	 * nfsv4 does some encoding while processing 
	 */
	nfserrp = rqstp->rq_res.head[0].iov_base
		+ rqstp->rq_res.head[0].iov_len;
	rqstp->rq_res.head[0].iov_len += sizeof(__be32);

	/* Now call the procedure handler, and encode NFS status. */
	nfserr = proc->pc_func(rqstp, rqstp->rq_argp, rqstp->rq_resp);
	nfserr = map_new_errors(rqstp->rq_vers, nfserr);
	if (nfserr == nfserr_dropit) {
		dprintk("nfsd: Dropping request; may be revisited later\n");
		nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
		return 0;
	}

	if (rqstp->rq_proc != 0)
		*nfserrp++ = nfserr;

	/* Encode result.
	 * For NFSv2, additional info is never returned in case of an error.
	 */
	if (!(nfserr && rqstp->rq_vers == 2)) {
		xdr = proc->pc_encode;
		if (xdr && !xdr(rqstp, nfserrp,
				rqstp->rq_resp)) {
			/* Failed to encode result. Release cache entry */
			dprintk("nfsd: failed to encode result!\n");
			nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
			*statp = rpc_system_err;
			return 1;
		}
	}

	/* Store reply in cache. */
	nfsd_cache_update(rqstp, proc->pc_cachetype, statp + 1);
	return 1;
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/fs/nfsd/nfssvc.c
	3	*
	4	* Central processing for nfsd.
	5	*
	6	* Authors: Olaf Kirch (okir@monad.swb.de)
	7	*
	8	* Copyright (C) 1995, 1996, 1997 Olaf Kirch <okir@monad.swb.de>
	9	*/
	10
1da177e4	11	#include <linux/module.h>
e8edc6e0	12	#include <linux/sched.h>
1da177e4 LT	13	#include <linux/time.h>
	14	#include <linux/errno.h>
	15	#include <linux/nfs.h>
	16	#include <linux/in.h>
	17	#include <linux/uio.h>
	18	#include <linux/unistd.h>
	19	#include <linux/slab.h>
	20	#include <linux/smp.h>
	21	#include <linux/smp_lock.h>
83144186	22	#include <linux/freezer.h>
1da177e4	23	#include <linux/fs_struct.h>
9867d76c	24	#include <linux/kthread.h>
1da177e4 LT	25
	26	#include <linux/sunrpc/types.h>
	27	#include <linux/sunrpc/stats.h>
	28	#include <linux/sunrpc/svc.h>
	29	#include <linux/sunrpc/svcsock.h>
	30	#include <linux/sunrpc/cache.h>
	31	#include <linux/nfsd/nfsd.h>
	32	#include <linux/nfsd/stats.h>
	33	#include <linux/nfsd/cache.h>
70c3b76c	34	#include <linux/nfsd/syscall.h>
1da177e4	35	#include <linux/lockd/bind.h>
a257cdd0	36	#include <linux/nfsacl.h>
1da177e4 LT	37
	38	#define NFSDDBG_FACILITY NFSDDBG_SVC
	39
	40	/* these signals will be delivered to an nfsd thread
	41	* when handling a request
	42	*/
	43	#define ALLOWED_SIGS (sigmask(SIGKILL))
	44	/* these signals will be delivered to an nfsd thread
	45	* when not handling a request. i.e. when waiting
	46	*/
	47	#define SHUTDOWN_SIGS (sigmask(SIGKILL) \| sigmask(SIGHUP) \| sigmask(SIGINT) \| sigmask(SIGQUIT))
1da177e4 LT	48
1da177e4 LT	49	extern struct svc_program nfsd_program;
9867d76c	50	static int nfsd(void *vrqstp);
1da177e4	51	struct timeval nfssvc_boot;
1da177e4 LT	52	static atomic_t nfsd_busy;
	53	static unsigned long nfsd_last_call;
	54	static DEFINE_SPINLOCK(nfsd_call_lock);
	55
bedbdd8b NB	56	/*
	57	* nfsd_mutex protects nfsd_serv -- both the pointer itself and the members
	58	* of the svc_serv struct. In particular, ->sv_nrthreads but also to some
	59	* extent ->sv_temp_socks and ->sv_permsocks. It also protects nfsdstats.th_cnt
	60	*
	61	* If (out side the lock) nfsd_serv is non-NULL, then it must point to a
	62	* properly initialised 'struct svc_serv' with ->sv_nrthreads > 0. That number
	63	* of nfsd threads must exist and each must listed in ->sp_all_threads in each
	64	* entry of ->sv_pools[].
	65	*
	66	* Transitions of the thread count between zero and non-zero are of particular
	67	* interest since the svc_serv needs to be created and initialized at that
	68	* point, or freed.
3dd98a3b JL	69	*
	70	* Finally, the nfsd_mutex also protects some of the global variables that are
	71	* accessed when nfsd starts and that are settable via the write_* routines in
	72	* nfsctl.c. In particular:
	73	*
	74	* user_recovery_dirname
	75	* user_lease_time
	76	* nfsd_versions
bedbdd8b NB	77	*/
	78	DEFINE_MUTEX(nfsd_mutex);
	79	struct svc_serv *nfsd_serv;
	80
3fb803a9 AG	81	#if defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL)
	82	static struct svc_stat nfsd_acl_svcstats;
	83	static struct svc_version * nfsd_acl_version[] = {
	84	[2] = &nfsd_acl_version2,
	85	[3] = &nfsd_acl_version3,
	86	};
	87
	88	#define NFSD_ACL_MINVERS 2
e8c96f8c	89	#define NFSD_ACL_NRVERS ARRAY_SIZE(nfsd_acl_version)
3fb803a9 AG	90	static struct svc_version *nfsd_acl_versions[NFSD_ACL_NRVERS];
	91
	92	static struct svc_program nfsd_acl_program = {
	93	.pg_prog = NFS_ACL_PROGRAM,
	94	.pg_nvers = NFSD_ACL_NRVERS,
	95	.pg_vers = nfsd_acl_versions,
1a8eff6d	96	.pg_name = "nfsacl",
3fb803a9 AG	97	.pg_class = "nfsd",
	98	.pg_stats = &nfsd_acl_svcstats,
	99	.pg_authenticate = &svc_set_client,
	100	};
	101
	102	static struct svc_stat nfsd_acl_svcstats = {
	103	.program = &nfsd_acl_program,
	104	};
	105	#endif /* defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL) */
	106
70c3b76c N	107	static struct svc_version * nfsd_version[] = {
	108	[2] = &nfsd_version2,
	109	#if defined(CONFIG_NFSD_V3)
	110	[3] = &nfsd_version3,
	111	#endif
	112	#if defined(CONFIG_NFSD_V4)
	113	[4] = &nfsd_version4,
	114	#endif
	115	};
	116
	117	#define NFSD_MINVERS 2
e8c96f8c	118	#define NFSD_NRVERS ARRAY_SIZE(nfsd_version)
70c3b76c N	119	static struct svc_version *nfsd_versions[NFSD_NRVERS];
	120
	121	struct svc_program nfsd_program = {
3fb803a9 AG	122	#if defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL)
	123	.pg_next = &nfsd_acl_program,
	124	#endif
70c3b76c N	125	.pg_prog = NFS_PROGRAM, /* program number */
	126	.pg_nvers = NFSD_NRVERS, /* nr of entries in nfsd_version */
	127	.pg_vers = nfsd_versions, /* version table */
	128	.pg_name = "nfsd", /* program name */
	129	.pg_class = "nfsd", /* authentication class */
	130	.pg_stats = &nfsd_svcstats, /* version table */
	131	.pg_authenticate = &svc_set_client, /* export authentication */
	132
	133	};
	134
6658d3a7 N	135	int nfsd_vers(int vers, enum vers_op change)
	136	{
	137	if (vers < NFSD_MINVERS \|\| vers >= NFSD_NRVERS)
	138	return -1;
	139	switch(change) {
	140	case NFSD_SET:
	141	nfsd_versions[vers] = nfsd_version[vers];
6658d3a7 N	142	#if defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL)
6658d3a7 N	143	if (vers < NFSD_ACL_NRVERS)
1a8eff6d	144	nfsd_acl_versions[vers] = nfsd_acl_version[vers];
6658d3a7	145	#endif
1a8eff6d	146	break;
6658d3a7 N	147	case NFSD_CLEAR:
	148	nfsd_versions[vers] = NULL;
	149	#if defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL)
	150	if (vers < NFSD_ACL_NRVERS)
1a8eff6d	151	nfsd_acl_versions[vers] = NULL;
6658d3a7 N	152	#endif
	153	break;
	154	case NFSD_TEST:
	155	return nfsd_versions[vers] != NULL;
	156	case NFSD_AVAIL:
	157	return nfsd_version[vers] != NULL;
	158	}
	159	return 0;
	160	}
1da177e4 LT	161	/*
	162	* Maximum number of nfsd processes
	163	*/
	164	#define NFSD_MAXSERVS 8192
	165
	166	int nfsd_nrthreads(void)
	167	{
	168	if (nfsd_serv == NULL)
	169	return 0;
	170	else
	171	return nfsd_serv->sv_nrthreads;
	172	}
	173
bc591ccf N	174	static void nfsd_last_thread(struct svc_serv *serv)
	175	{
	176	/* When last nfsd thread exits we need to do some clean-up */
7a182083 TT	177	struct svc_xprt *xprt;
7a182083 TT	178	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list)
24e36663	179	lockd_down();
bc591ccf N	180	nfsd_serv = NULL;
	181	nfsd_racache_shutdown();
	182	nfs4_state_shutdown();
	183
e096bbc6 JL	184	printk(KERN_WARNING "nfsd: last server has exited, flushing export "
	185	"cache\n");
	186	nfsd_export_flush();
bc591ccf	187	}
6658d3a7 N	188
	189	void nfsd_reset_versions(void)
	190	{
	191	int found_one = 0;
	192	int i;
	193
	194	for (i = NFSD_MINVERS; i < NFSD_NRVERS; i++) {
	195	if (nfsd_program.pg_vers[i])
	196	found_one = 1;
	197	}
	198
	199	if (!found_one) {
	200	for (i = NFSD_MINVERS; i < NFSD_NRVERS; i++)
	201	nfsd_program.pg_vers[i] = nfsd_version[i];
	202	#if defined(CONFIG_NFSD_V2_ACL) \|\| defined(CONFIG_NFSD_V3_ACL)
	203	for (i = NFSD_ACL_MINVERS; i < NFSD_ACL_NRVERS; i++)
	204	nfsd_acl_program.pg_vers[i] =
	205	nfsd_acl_version[i];
	206	#endif
	207	}
	208	}
	209
bedbdd8b	210
b41b66d6	211	int nfsd_create_serv(void)
02a375f0 N	212	{
02a375f0 N	213	int err = 0;
bedbdd8b NB	214
bedbdd8b NB	215	WARN_ON(!mutex_is_locked(&nfsd_mutex));
02a375f0	216	if (nfsd_serv) {
9a24ab57	217	svc_get(nfsd_serv);
02a375f0 N	218	return 0;
02a375f0 N	219	}
596bbe53 N	220	if (nfsd_max_blksize == 0) {
	221	/* choose a suitable default */
	222	struct sysinfo i;
	223	si_meminfo(&i);
	224	/* Aim for 1/4096 of memory per thread
	225	* This gives 1MB on 4Gig machines
	226	* But only uses 32K on 128M machines.
	227	* Bottom out at 8K on 32M and smaller.
	228	* Of course, this is only a default.
	229	*/
	230	nfsd_max_blksize = NFSSVC_MAXBLKSIZE;
44c55600	231	i.totalram <<= PAGE_SHIFT - 12;
596bbe53 N	232	while (nfsd_max_blksize > i.totalram &&
	233	nfsd_max_blksize >= 810242)
	234	nfsd_max_blksize /= 2;
	235	}
02a375f0 N	236
02a375f0 N	237	atomic_set(&nfsd_busy, 0);
e096bbc6	238	nfsd_serv = svc_create_pooled(&nfsd_program, nfsd_max_blksize,
a75c5d01	239	nfsd_last_thread, nfsd, THIS_MODULE);
02a375f0 N	240	if (nfsd_serv == NULL)
02a375f0 N	241	err = -ENOMEM;
bedbdd8b	242
02a375f0 N	243	do_gettimeofday(&nfssvc_boot); /* record boot time */
	244	return err;
	245	}
	246
	247	static int nfsd_init_socks(int port)
	248	{
	249	int error;
	250	if (!list_empty(&nfsd_serv->sv_permsocks))
	251	return 0;
	252
02a375f0	253	error = lockd_up(IPPROTO_UDP);
4a3ae42d	254	if (error >= 0) {
d7c9f1ed	255	error = svc_create_xprt(nfsd_serv, "udp", port,
482fb94e	256	SVC_SOCK_DEFAULTS);
4a3ae42d N	257	if (error < 0)
	258	lockd_down();
	259	}
02a375f0 N	260	if (error < 0)
	261	return error;
	262
02a375f0	263	error = lockd_up(IPPROTO_TCP);
4a3ae42d	264	if (error >= 0) {
d7c9f1ed	265	error = svc_create_xprt(nfsd_serv, "tcp", port,
482fb94e	266	SVC_SOCK_DEFAULTS);
4a3ae42d N	267	if (error < 0)
	268	lockd_down();
	269	}
02a375f0 N	270	if (error < 0)
02a375f0 N	271	return error;
02a375f0 N	272	return 0;
	273	}
	274
eed2965a GB	275	int nfsd_nrpools(void)
	276	{
	277	if (nfsd_serv == NULL)
	278	return 0;
	279	else
	280	return nfsd_serv->sv_nrpools;
	281	}
	282
	283	int nfsd_get_nrthreads(int n, int *nthreads)
	284	{
	285	int i = 0;
	286
	287	if (nfsd_serv != NULL) {
	288	for (i = 0; i < nfsd_serv->sv_nrpools && i < n; i++)
	289	nthreads[i] = nfsd_serv->sv_pools[i].sp_nrthreads;
	290	}
	291
	292	return 0;
	293	}
	294
	295	int nfsd_set_nrthreads(int n, int *nthreads)
	296	{
	297	int i = 0;
	298	int tot = 0;
	299	int err = 0;
	300
bedbdd8b NB	301	WARN_ON(!mutex_is_locked(&nfsd_mutex));
bedbdd8b NB	302
eed2965a GB	303	if (nfsd_serv == NULL \|\| n <= 0)
	304	return 0;
	305
	306	if (n > nfsd_serv->sv_nrpools)
	307	n = nfsd_serv->sv_nrpools;
	308
	309	/* enforce a global maximum number of threads */
	310	tot = 0;
	311	for (i = 0; i < n; i++) {
	312	if (nthreads[i] > NFSD_MAXSERVS)
	313	nthreads[i] = NFSD_MAXSERVS;
	314	tot += nthreads[i];
	315	}
	316	if (tot > NFSD_MAXSERVS) {
	317	/* total too large: scale down requested numbers */
	318	for (i = 0; i < n && tot > 0; i++) {
	319	int new = nthreads[i] * NFSD_MAXSERVS / tot;
	320	tot -= (nthreads[i] - new);
	321	nthreads[i] = new;
	322	}
	323	for (i = 0; i < n && tot > 0; i++) {
	324	nthreads[i]--;
	325	tot--;
	326	}
	327	}
	328
	329	/*
	330	* There must always be a thread in pool 0; the admin
	331	* can't shut down NFS completely using pool_threads.
	332	*/
	333	if (nthreads[0] == 0)
	334	nthreads[0] = 1;
	335
	336	/* apply the new numbers */
eed2965a GB	337	svc_get(nfsd_serv);
	338	for (i = 0; i < n; i++) {
	339	err = svc_set_num_threads(nfsd_serv, &nfsd_serv->sv_pools[i],
	340	nthreads[i]);
	341	if (err)
	342	break;
	343	}
	344	svc_destroy(nfsd_serv);
eed2965a GB	345
	346	return err;
	347	}
	348
1da177e4 LT	349	int
	350	nfsd_svc(unsigned short port, int nrservs)
	351	{
	352	int error;
bedbdd8b NB	353
bedbdd8b NB	354	mutex_lock(&nfsd_mutex);
6658d3a7	355	dprintk("nfsd: creating service\n");
1da177e4 LT	356	error = -EINVAL;
	357	if (nrservs <= 0)
	358	nrservs = 0;
	359	if (nrservs > NFSD_MAXSERVS)
	360	nrservs = NFSD_MAXSERVS;
	361
	362	/* Readahead param cache - will no-op if it already exists */
	363	error = nfsd_racache_init(2*nrservs);
	364	if (error<0)
	365	goto out;
e8ff2a84	366	nfs4_state_start();
02a375f0 N	367
	368	nfsd_reset_versions();
	369
	370	error = nfsd_create_serv();
	371
	372	if (error)
	373	goto out;
	374	error = nfsd_init_socks(port);
	375	if (error)
	376	goto failure;
	377
eec09661	378	error = svc_set_num_threads(nfsd_serv, NULL, nrservs);
1da177e4	379	failure:
1da177e4	380	svc_destroy(nfsd_serv); /* Release server */
1da177e4	381	out:
bedbdd8b	382	mutex_unlock(&nfsd_mutex);
1da177e4 LT	383	return error;
	384	}
	385
	386	static inline void
	387	update_thread_usage(int busy_threads)
	388	{
	389	unsigned long prev_call;
	390	unsigned long diff;
	391	int decile;
	392
	393	spin_lock(&nfsd_call_lock);
	394	prev_call = nfsd_last_call;
	395	nfsd_last_call = jiffies;
	396	decile = busy_threads*10/nfsdstats.th_cnt;
	397	if (decile>0 && decile <= 10) {
	398	diff = nfsd_last_call - prev_call;
	399	if ( (nfsdstats.th_usage[decile-1] += diff) >= NFSD_USAGE_WRAP)
	400	nfsdstats.th_usage[decile-1] -= NFSD_USAGE_WRAP;
	401	if (decile == 10)
	402	nfsdstats.th_fullcnt++;
	403	}
	404	spin_unlock(&nfsd_call_lock);
	405	}
	406
	407	/*
	408	* This is the NFS server kernel thread
	409	*/
9867d76c JL	410	static int
9867d76c JL	411	nfsd(void *vrqstp)
1da177e4	412	{
9867d76c	413	struct svc_rqst rqstp = (struct svc_rqst ) vrqstp;
1da177e4	414	struct fs_struct *fsp;
1da177e4	415	sigset_t shutdown_mask, allowed_mask;
9867d76c JL	416	int err, preverr = 0;
9867d76c JL	417	unsigned int signo;
1da177e4 LT	418
1da177e4 LT	419	/* Lock module and set up kernel thread */
bedbdd8b	420	mutex_lock(&nfsd_mutex);
1da177e4	421
9867d76c	422	/* At this point, the thread shares current->fs
1da177e4 LT	423	* with the init process. We need to create files with a
	424	* umask of 0 instead of init's umask. */
	425	fsp = copy_fs_struct(current->fs);
	426	if (!fsp) {
	427	printk("Unable to start nfsd thread: out of memory\n");
	428	goto out;
	429	}
	430	exit_fs(current);
	431	current->fs = fsp;
	432	current->fs->umask = 0;
	433
	434	siginitsetinv(&shutdown_mask, SHUTDOWN_SIGS);
	435	siginitsetinv(&allowed_mask, ALLOWED_SIGS);
	436
9867d76c JL	437	/*
	438	* thread is spawned with all signals set to SIG_IGN, re-enable
	439	* the ones that matter
	440	*/
	441	for (signo = 1; signo <= _NSIG; signo++) {
	442	if (!sigismember(&shutdown_mask, signo))
	443	allow_signal(signo);
	444	}
bedbdd8b	445
1da177e4	446	nfsdstats.th_cnt++;
bedbdd8b NB	447	mutex_unlock(&nfsd_mutex);
bedbdd8b NB	448
1da177e4 LT	449	/*
	450	* We want less throttling in balance_dirty_pages() so that nfs to
	451	* localhost doesn't cause nfsd to lock up due to all the client's
	452	* dirty pages.
	453	*/
	454	current->flags \|= PF_LESS_THROTTLE;
83144186	455	set_freezable();
1da177e4 LT	456
	457	/*
	458	* The main request loop
	459	*/
	460	for (;;) {
	461	/* Block all but the shutdown signals */
	462	sigprocmask(SIG_SETMASK, &shutdown_mask, NULL);
	463
	464	/*
	465	* Find a socket with data available and call its
	466	* recvfrom routine.
	467	*/
6fb2b47f	468	while ((err = svc_recv(rqstp, 6060HZ)) == -EAGAIN)
1da177e4	469	;
9867d76c	470	if (err == -EINTR)
1da177e4	471	break;
9867d76c JL	472	else if (err < 0) {
	473	if (err != preverr) {
	474	printk(KERN_WARNING "%s: unexpected error "
	475	"from svc_recv (%d)\n", __func__, -err);
	476	preverr = err;
	477	}
	478	schedule_timeout_uninterruptible(HZ);
	479	continue;
	480	}
	481
1da177e4 LT	482	update_thread_usage(atomic_read(&nfsd_busy));
	483	atomic_inc(&nfsd_busy);
	484
	485	/* Lock the export hash tables for reading. */
	486	exp_readlock();
	487
9867d76c	488	/* Process request with signals blocked. */
1da177e4 LT	489	sigprocmask(SIG_SETMASK, &allowed_mask, NULL);
1da177e4 LT	490
6fb2b47f	491	svc_process(rqstp);
1da177e4 LT	492
	493	/* Unlock export hash tables */
	494	exp_readunlock();
	495	update_thread_usage(atomic_read(&nfsd_busy));
	496	atomic_dec(&nfsd_busy);
	497	}
	498
24e36663	499	/* Clear signals before calling svc_exit_thread() */
9e416052	500	flush_signals(current);
1da177e4	501
bedbdd8b	502	mutex_lock(&nfsd_mutex);
1da177e4 LT	503	nfsdstats.th_cnt --;
	504
	505	out:
	506	/* Release the thread */
	507	svc_exit_thread(rqstp);
	508
	509	/* Release module */
bedbdd8b	510	mutex_unlock(&nfsd_mutex);
1da177e4	511	module_put_and_exit(0);
9867d76c	512	return 0;
1da177e4 LT	513	}
1da177e4 LT	514
32c1eb0c AA	515	static __be32 map_new_errors(u32 vers, __be32 nfserr)
	516	{
	517	if (nfserr == nfserr_jukebox && vers == 2)
	518	return nfserr_dropit;
	519	if (nfserr == nfserr_wrongsec && vers < 4)
	520	return nfserr_acces;
	521	return nfserr;
	522	}
	523
1da177e4	524	int
c7afef1f	525	nfsd_dispatch(struct svc_rqst rqstp, __be32 statp)
1da177e4 LT	526	{
	527	struct svc_procedure *proc;
	528	kxdrproc_t xdr;
ad451d38 AV	529	__be32 nfserr;
ad451d38 AV	530	__be32 *nfserrp;
1da177e4 LT	531
	532	dprintk("nfsd_dispatch: vers %d proc %d\n",
	533	rqstp->rq_vers, rqstp->rq_proc);
	534	proc = rqstp->rq_procinfo;
	535
	536	/* Check whether we have this call in the cache. */
	537	switch (nfsd_cache_lookup(rqstp, proc->pc_cachetype)) {
	538	case RC_INTR:
	539	case RC_DROPIT:
	540	return 0;
	541	case RC_REPLY:
	542	return 1;
	543	case RC_DOIT:;
	544	/* do it */
	545	}
	546
	547	/* Decode arguments */
	548	xdr = proc->pc_decode;
ad451d38	549	if (xdr && !xdr(rqstp, (__be32*)rqstp->rq_arg.head[0].iov_base,
1da177e4 LT	550	rqstp->rq_argp)) {
	551	dprintk("nfsd: failed to decode arguments!\n");
	552	nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
	553	*statp = rpc_garbage_args;
	554	return 1;
	555	}
	556
	557	/* need to grab the location to store the status, as
	558	* nfsv4 does some encoding while processing
	559	*/
	560	nfserrp = rqstp->rq_res.head[0].iov_base
	561	+ rqstp->rq_res.head[0].iov_len;
ad451d38	562	rqstp->rq_res.head[0].iov_len += sizeof(__be32);
1da177e4 LT	563
	564	/* Now call the procedure handler, and encode NFS status. */
	565	nfserr = proc->pc_func(rqstp, rqstp->rq_argp, rqstp->rq_resp);
32c1eb0c	566	nfserr = map_new_errors(rqstp->rq_vers, nfserr);
1da177e4	567	if (nfserr == nfserr_dropit) {
45457e09	568	dprintk("nfsd: Dropping request; may be revisited later\n");
1da177e4 LT	569	nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
	570	return 0;
	571	}
	572
	573	if (rqstp->rq_proc != 0)
	574	*nfserrp++ = nfserr;
	575
	576	/* Encode result.
	577	* For NFSv2, additional info is never returned in case of an error.
	578	*/
	579	if (!(nfserr && rqstp->rq_vers == 2)) {
	580	xdr = proc->pc_encode;
	581	if (xdr && !xdr(rqstp, nfserrp,
	582	rqstp->rq_resp)) {
	583	/* Failed to encode result. Release cache entry */
	584	dprintk("nfsd: failed to encode result!\n");
	585	nfsd_cache_update(rqstp, RC_NOCACHE, NULL);
	586	*statp = rpc_system_err;
	587	return 1;
	588	}
	589	}
	590
	591	/* Store reply in cache. */
	592	nfsd_cache_update(rqstp, proc->pc_cachetype, statp + 1);
	593	return 1;
	594	}